We studied the first 100 consecutive patients (125 knees) to undergo knee arthroplasty with the original total condylar-I prosthesis for either osteoarthritis or osteonecrosis on the Knee Service at The Hospital for Special Surgery. They were followed for an average of 6.6 years (range, five to nine years), which is the longest follow-up of a total knee prosthesis that is still widely used in its original and unmodified form. At five to nine years of follow-up, eleven patients (fourteen knees) were dead and ten patients (eleven knees) could not be traced, leaving 100 knees (seventy-nine patients) for clinical evaluation. This included one knee from which the prosthesis was removed for sepsis at six months, four that were revised for pain, and two that were revised because of subluxation. Two patients had a loose tibial component. Sixty-four knees were rated as excellent and twenty-seven were rated as good. There were two fair and seven poor results. Detailed radiographic evaluation of the 100 knees showed that only minimum changes in the varus or valgus alignment or the fixation of the prosthetic components had occurred with time. At follow-up, 41 per cent of the tibial components showed some evidence of radiolucency, although with the exception of the two loose components just mentioned there were no complete or circumferential radiolucent lines. Most radiolucencies were present by the end of the first postoperative year, and in 81 per cent of the knees there was no further change. Although a special attempt was made to determine any evidence of wear of the polyethylene tibial component, we were unable to demonstrate unequivocal wear in any knee. Our experience had led us to conclude that the long-term success of current total knee designs will probably not be significantly affected by polyethylene wear or loss of component fixation. The majority of failures in our series were attributed to errors in surgical technique.